Tests fail when run with miri #15
Comments
|
Here's a simpler test case: let mut m = LinkedList::new();
m.push_back(2);
m.push_back(3);
assert_eq!(m.pop_front(), Some(2));which fails with: |
|
A further simplification: let ptr: *mut Node<_>;
let mut head;
{
let mut node = Box::new(Node::new(2));
// unconditionally make the new node the new tail
ptr = &mut *node;
head = Some(node);
}
{
let mut node = Box::new(Node::new(3));
let tail = unsafe { &mut *(ptr as *mut Node<_>) };
node.prev = Raw::some(tail);
tail.next = Some(node);
}
head.take(); |
|
And here's the essence of what's going on: let mut node = Box::new(Node::new(2));
let mut n2 = Box::new(Node::new(3));
let ptr: *mut Node<_> = &mut *node;
let tail = unsafe { &mut *(ptr as *mut Node<_>) };
let mut head = Some(node);
tail.next = Some(n2);If we swap the order of It's not clear to me why miri's is complaining about this: |
|
Can you make this into a self-contained example for the playground? That'd make analysis much easier for me. |
|
Yep. Will do. |
|
|
|
It seems like the use of Box in this crate basically needs to go. It looks like this happened in std's LinkedList in rust-lang/rust#34608 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The text was updated successfully, but these errors were encountered: